AI-Supported Work Reconfiguration and Emerging Inbound Open Innovation Practices: An Exploratory Multiple-Case Study of Cross-Border E-Commerce SMEs

Artificial intelligence (AI) is changing how firms search for, interpret, and act on externally generated knowledge in digital innovation ecosystems. Existing research motivates closer examination of how resource-constrained SMEs organize the internal workflows through which AI-supported customer, market, competitor, and platform information becomes usable for innovation. This study examines how cross-border e-commerce SMEs organize AI-supported workflows and team roles to handle external knowledge for emerging inbound open innovation. The empirical design is an exploratory, theory-elaborating qualitative multiple-case study of two contrasting SMEs. The analysis draws on ten face-to-face semi-structured interviews across strategic, managerial, technical, and frontline roles, totaling 183 min 32 s (approximately 184 recorded minutes). An abductive thematic case-analysis approach combines data-near coding, within-case analysis, cross-case synthesis, and iteration with relevant theory. The analysis suggests an analytically ordered framework rather than a verified longitudinal sequence. Case A exhibits experiment-led diffusion, whereas Case B exhibits technical-partner-led workflow design. Across the cases, participants described modular human–AI–human sequences, broader task integration, changing feedback arrangements, and greater emphasis on human review. These reported arrangements are associated with three emerging capability-related practice dimensions: external knowledge sensing, knowledge recombination, and agile implementation. Disconfirming accounts show that the patterns depend on task–AI fit and that overreliance can weaken independent market judgment. The study contributes to research on AI-enabled inbound open innovation by showing that access to AI and platform data is insufficient: external inputs become actionable when SMEs reorganize workflows and roles around prompting, interpretation, recombination, and human accountability. It also elaborates how human–AI collaboration operates as an internal microfoundation of customer-centric, platform-mediated openness. For managers, the findings suggest moving from isolated tool use toward capability-oriented workflow design while retaining human review. The conclusions offer contextualized theoretical insight rather than statistical generalization.